Update JinaAI rerank model token limit in rerankerWindowSize method

Author: Jan-Kazlouski-elastic

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/services/openshiftai/OpenShiftAiService.java

@@ -342,7 +342,7 @@ private OpenShiftAiModel createModelFromPersistent(
     @Override
     public int rerankerWindowSize(String modelId) {
         // OpenShift AI uses Cohere and JinaAI rerank protocols for reranking
-        // JinaAI rerank model has 8000 tokens limit length https://jina.ai/models/jina-reranker-v2-base-multilingual
+        // JinaAI rerank model has 131K tokens limit https://jina.ai/models/jina-reranker-v3/
         // Cohere rerank model truncates at 4096 tokens https://docs.cohere.com/reference/rerank
         // We choose a conservative limit based on these two models
         // Using 1 token = 0.75 words as a rough estimate, we get 3072 words allowing for some headroom, we set the window size below 3072